One person in the treatment group, to the left in Figure 11, shows a reduction of total WOMAC score by over 50. This was a 57-year-old woman with no hip OA but bilateral hip pain < 6 months but more than 3 months. She was married, lived in a small village and worked at the old people’s home. Reporting on what the Hip School had meant to her, she wrote, "I have learnt how important it is to walk, stand, sit, correctly; move your legs while sitting. How important it is to stand up after sitting for a time. To consider your problems in a more positive way, to learn how to handle the disease".
Passive hip range of motion
Of the 168 patients in papers III and IV, 50 had no hip OA, 77 had unilateral hip OA and 41 had bilateral hip OA according to the radiological reports (Table 1). As there were age and gender differences between OA patients and non-OA patients (Table 2), logistic regression analyses were made. Table 11 demonstrates the relationships between radiological signs of hip OA, internal rotation < 15o and flexion ≤ 115o for the whole group (n = 336) as dependent variables, and the independent variables of age, sex and OA or non-OA. Being male
contributed almost as much as having OA to the risk of having internal rotation < 15o.
Mean PROM differed in all directions between hips with OA and those without (Figure 12), and also between the sexes. For hips in patients 65 years or older, there were significant differences in all directions, except adduction and external rotation, compared to younger patients. Differences in male and female proportions and age between patients with and without OA (Table 2) prompted analyses for OA and non-OA hips for males and females, and older and younger, separately. These showed the same differences between OA and non-OA.
Table 11. Logistic regression analyses for flexion ≥ 116o or ≤ 115o and internal rotation ≥ 15o or < 15o for the whole group (n = 336) as dependent variables, and age, sex, osteoarthritis (OA) or non-OA as independent variables. Sex and OA or non-OA was entered into the model as category variables.
150 100 50 00 50 100 150
0 50 100 150 200 250 300
Extension Internal rot.
Adduction
Flexion External rot.
Abduction non OA mean n=177 OA mean n=159 1 sd
Degrees
Degrees Total PROM
Figure 12. Means for passive range of motion (PROM) and standard deviations (sd), for osteoarthritic (OA) hips (n = 159) and non-OA hips (n = 177), for the six different directions tested and for total PROM.
Among patients with unilateral hip OA (n = 77), 14 (18%) had decreased PROM in all directions in the OA hip compared to the non-OA hip. The other 63 (82%) patients had equal or more PROM in the OA hip than in the non-OA hip in between 1 to 5 directions (mean 2).
Equal or more PROM in the OA hip was most common in extension (31 cases), followed by external rotation and abduction; and less common in internal rotation (13 cases).
PROM and capsular patterns – paper III
For OA hips (n = 159) there were 68, 70 and 129 different PROM patterns, depending on PROM norm used, and for non-OA 68 and 138 (n = 177). The proportion of hips with limitations in all six directions varied between 0 and 21%, depending on the PROM norm used.
Tests of Cyriax’s capsular pattern
Five OA hips (3%) and 1 non-OA hip (<1%) had internal rotation followed by flexion as the two most limited directions when the PROM from the symptom-free hips was used as norm and five OA hips (3%) when Kaltenborn’s PROM norms were used.
For unilateral hip OA with PROM in the uninvolved hip as norm, nine hips (12%) had most limitations in internal rotation followed by flexion and 1 hip (1%) had most limitations in flexion, abduction or internal rotation followed by extension (Figure 8).
Limitations in abduction followed by flexion and internal rotation, irrespective of the ordering between them, occurred in 11 OA hips (7%) and 1 non-OA hip (<1%) when using the PROM norms from the symptom-free hips. When using Kaltenborn’s PROM norms, 33 OA hips (21%) and 18 non-OA (10%) were found, and seven hips (9%) when using the uninvolved side as PROM norm.
Test of Kaltenborn's capsular pattern
One non-OA hip (<1%) had limited internal rotation and extension, with the PROM in the symptom-free hips as norms and one OA hip (1%) with the uninvolved side as norm (Figure 8).
Diagnostic tests of the ACR clinical criteria – paper IV
Among the 212 hips with self-reported hip pain in paper IV, the prevalence of radiological signs of hip OA was 0.64 (n = 135). When comparing the working version of the ACR clinical criteria and the radiological reports, 115 hips had OA with both classification
systems, giving a sensitivity of 0.85 and a specificity of 0.25. Twenty hips with self-reported hip pain had radiological signs of hip OA but had non-ACR OA (Table 8). The positive predictive value was 0.66 and the negative predictive value 0.53. Percentage agreement was 0.63 and, when corrected for chance, 0.11, which is to be considered poor.
Totally 78 hips (37%) were classified with disagreement: twenty-three of the male hips (29
%) and 55 of the female (42%). The combination of passive internal rotation ≥ 15o and age >
50 years gave most of the disagreements for both male and female hips (n = 56).
HOOS – the extension of the WOMAC – paper V Four criteria for the item selection process
Criterion A was test-retest agreement of the scorings of each item of 60% or more. All items were examined with regard to percentage of patients reporting best possible scores (no pain, no stiffness, no activity limitations, never or not at all), making it impossible to show improvement over time when using e.g. a responder criterion to categorize an individual’s response to treatment (Dougados et al. 2000). Thus criterion B was whether an item scored minimum (0) in fewer than 30% of the cases. The mean score of the WOMAC items was 1.2.
Items with higher mean scores (more symptoms) would increase sub-scale scores in the new instrument and further permit ability to detect change in patients, on a group basis, over time.
Criterion C was selected as a mean score higher than 1.2 for each item. Finally the
importance of each item was rated with three alternatives: unimportant, somewhat important or very important (scored 1, 2 or 3); not, however, intended to be a part of the HOOS LK 1.1 instrument. Criterion D was met if fewer than 20% of the subjects scored the item as
‘unimportant’. Based on the results, ten of the 18 KOOS and all five additional items were included in the HOOS LK 1.1 (reported in appendices B and C).
Test of the new questionnaire
HOOS LK 1.1 was established with 39 items, in five sub-scales, Pain (P), Symptoms (other, including stiffness) (S), Activity limitations – daily living (A), Activity limitations – sport and recreation (SP) and Hip-related Quality of life (Q). To enhance the interpretation, all sub-scale scores were transformed in the data analysis to 0-100, best-to-worst. For user’s guide see appendix A.
The intra-class correlation coefficients (ICC 2,1) for the HOOS sub-scales were for Pain 0.89, Symptoms (other) 0.86, Activity limitations – daily living 0.89, Activity limitations – sport and recreation 0.91 and Hip-related Quality of life 0.78, which must be considered high. On the first test occasion the Cronbach’s alphas were as follows: Pain 0.93, Symptoms 0.82, Activity limitations – daily living 0.96, Activity limitations – sport and recreation 0.88 and Hip-related Quality of life 0.77, which must be considered acceptable.
In HOOS, 38% of the items had a floor effect (Criterion B) compared to 58% in the
WOMAC. Three of the HOOS sub-scales showed higher medians (more symptoms) than the WOMAC did, with the highest median in the newly-added Hip-related Quality of life and Activity limitations – sport and recreation, where median scores were 44 and 41, respectively (Table 12). The items in the Symptoms sub-scale are very disparate. Half our patients
assessed zero in one or more of the new items. Due to the mathematical procedure when transforming the scale scores to 0-100, many zero assessments resulted in lowered sub-scale scores (less symptoms).
One single component was found when the items in the five sub-scales, one at a time, were entered in within-scale principal component analysis with varimax rotation, explaining between 59 and 73% of the scale score variance. There was one exception: the sub-scale Activity limitations – daily living loaded on two components, with the first explaining 59% of the variance and the other 8%. Varimax rotation revealed that the first comprised activities with heavy loading and maintenance of postures while the second comprised mostly activities that required a good ROM.
Comparison of WOMAC scores
In table 10 the sub-scale scores of WOMAC from papers I, II and V are presented together with p-values (Wilcoxon matched-pairs signed-rank test), effect sizes and standardised response means. The mean scores in the treatment group before Hip School were not high, which gave low effect sizes and SRM. As seen in Table 12 three of the HOOS sub-scales showed higher means than the WOMAC did. For mathematical reasons, this improves the HOOS’s ability to detect clinically important change over time.
Table 12. Comparison of sub-scale scores of Western Ontario and
McMaster Universities Osteoarthritis Index (WOMAC) and Hip disability and osteoarthritis outcome score (HOOS), paper V.
Comparison WOMAC and HOOS Paper V
sub-scale scores WOMAC HOOS
n 52
Pain
- Mean 26 33
- SD 17 20
- Median 25 31
- Interquartile range 24 33
- P-value 0.000
Symptoms (other including stiffness)
- Mean 39 37
- SD 21 21
- Median 38 35
- Interquartile range 25 30
- P-value 0.404
Activity limitations – daily living - Mean
- SD - Median
- Interquartile range - P-value
30 20 28 35 -Activity limitations – sport and
recreation
- Mean - 43
- SD - 27
- Median - 41
- Interquartile range - 47
- P-value
-Quality of life – hip related
- Mean - 45
- SD - 20
- Median - 44
- Interquartile range - 31
- P-value
-SD: standard deviation